Decentralized auto-scaling of network architectures
Disclosed herein are systems, devices, and methods for providing auto-scaling in a cluster of device instances. In one embodiment, a method is disclosed comprising updating, using a distributed counter, a metric associated with one or more instances executing a network application; identifying that the metric has exceeded a threshold defined in a scaling policy based on comparing the distributed counter to the scaling policy; identifying a command to execute in response to the metric exceeding the threshold; and executing the command to modify the one or more instances.
1. A device comprising:
a processor; and
a storage medium for tangibly storing thereon program logic for execution by the processor, the stored program logic comprising:
logic, executed by the processor, for executing a distributed counter to update a count associated with a respective metric,
logic, executed by the processor, for transmitting a first value of the count to one or more instances, the one or more instances executing respective distributed counters,
logic, executed by the processor, for receiving a second value of the count from the one or more instances,
logic, executed by the processor, for updating, using the distributed counter, a metric associated with the one or more instances based on the second value,
logic, executed by the processor, for identifying that the metric has exceeded a threshold defined in a scaling policy, and
logic, executed by the processor, for executing a command in response to the metric exceeding the threshold.
2. The device of claim 1 the logic for executing a command in response to the metric exceeding the threshold comprising logic, executed by the processor, for executing a command selected from the group consisting of instantiating a new instance, destroying one of the one or more instances, and modifying one of the one or more instances.
3. The device of claim 1 , the first value causing the respective distributed counter of the one or more instances to update a metric associated with the distributed counter executing on the device.
4. The device of claim 1 , the logic for executing the command comprising transmitting an alert to a scaling component.
5. The device of claim 3 , the executing the command comprising issuing the command from a master instance to an image management infrastructure component.
6. The device of claim 1 , the logic for updating a metric associated with one or more instances comprising:
logic, executed by the processor, for performing an operation;
logic, executed by the processor, for updating a local counter corresponding to the metric; and
logic, executed by the processor, for sending a message to at least one other instance, the message causing the at least one other instance to update a distributed count value.
7. A system comprising:
one or more instances executing a network application, the one or more instances configured to:
update, using a distributed counter stored locally by each of the one or more instances, a metric associated with each of the one or more instances; and
configure one of the one or more instances as a scaling component, the scaling component configured to:
identify that the metric has exceeded a threshold defined in a scaling policy based on comparing the distributed counter to a value in the scaling policy, and
execute a command to modify the one or more instances in response to the metric exceeding the threshold.
8. The system of claim 7 , the scaling component further configured to pull the metric from the one or more instances.
9. The system of claim 8 , the scaling component further configured to retrieve the scaling policy stored locally on the scaling component and identify a threshold in the scaling policy.
10. The system of claim 8 the scaling component further configured to execute a command selected from the group consisting of instantiating a new instance, destroying one of the one or more instances, and modifying one of the one or more instances.
11. The system of claim 8 , the one or more instances further configured to:
perform, by a selected instance from the one or more instances, an operation;
update, by the selected instance, a local counter corresponding to the metric; and
send, by the selected instance, a message to at least one other instance, the message causing the at least one other instance to update a distributed count value.
12. A method comprising:
receiving, by one or more instances executing a network application, requests from a client device;
updating, using distributed counters implemented by the one or more instances and stored locally by each of the one or more instances, a metric associated with each of the one or more instances based on the requests;
identifying, by a master instance in the one or more instances, that the metric has exceeded a threshold defined in a scaling policy based on comparing the distributed counter to the scaling policy;
identifying, by the master instance, a command to execute in response to the metric exceeding the threshold; and
executing, by the master instance, the command to modify the one or more instances.
13. The method of claim 12 , the executing the command to modify the one or more instances comprising executing a command selected from the group consisting of instantiating a new instance, destroying one of the one or more instances, and modifying one of the one or more instances.
14. The method of claim 12 , the updating a metric associated with one or more instances executing a network application comprising:
performing, by a selected instance from the one or more instances, an operation;
updating, by the selected instance, a local counter corresponding to the metric; and
sending, by the selected instance, a message to at least one other instance, the message causing the at least one other instance to update a distributed count value.
15. The method of claim 12 , further comprising periodically selecting a master instance from the one or more instances.
16. The method of claim 15 , the periodically selecting a master instance from the one or more instances performed using a selection algorithm selected from the group consisting of a fixed selection, round robin algorithm, probabilistic selection algorithm, or election periodic timer algorithm.
17. The method of claim 15 , the identifying that the metric has exceeded a threshold defined in a scaling policy comprising retrieving, by the master instance, the scaling policy stored locally on the master instance and identifying a threshold in the scaling policy.
18. The method of claim 15 , the identifying a command to execute in response to the metric exceeding the threshold comprising issuing the command from the master instance to an image management infrastructure component.
19. The method of claim 12 , the identifying that the metric has exceeded a threshold defined in a scaling policy comprising pulling, by a scaling component, the metric from the one or more instances.
20. The method of claim 19 , the identifying that the metric has exceeded a threshold defined in a scaling policy comprising retrieving, by the scaling component, the scaling policy stored locally on the scaling component and identifying a threshold in the scaling policy.